Periodic Reporting for period 2 - COSMOS (DevOps for Complex Cyber-physical Systems)
Período documentado: 2022-07-01 hasta 2024-03-31
Cyber-physical Systems (CPS) – from robotics to transportation to medical devices – will play a crucial role in the quality of life of European citizens and the future of the European economy. One specific example is the public transport system where the percentage of fully automated operation is expected to increase from today’s 30% to 70% by 2030: CPS is a key enabler in this and many other industrial and societal evolutions. Increasing automation to such an extent, however, gives rise to many challenges, at the crux of which lies the hardware and software symbiosis. Several new challenges arise from the increasing complexity of CPS software, seamless connectivity, abundant compute power, and hardware heterogeneity. Emerging CPS are characterised by an evolving development that never ends, and engineering practitioners in the field are facing fundamental development challenges: observability, testability and predictability of the behaviour of emerging CPS is highly limited and, unfortunately, their usage in the real world can lead to accidents, sometimes tragically also involving humans. DevOps practices and tools are potentially the right solution to this problem, but they have not been developed to be applied in CPS domains. COSMOS takes on these challenges by delivering the technological and methodological advances necessary to enable DevOps for development of complex, trustworthy and reliable CPS solutions.
CHALLENGE
A key challenge in applying DevOps practices to CPS domains is that it requires specific development and verification strategies able to include Hardware-in-the-Loop (HiL) capabilities. Also, embedded systems design, manufacturing, and testing have different, longer lead times and cycle times than enterprise software, leading to longer V&V procedures and higher testing costs (typically over 25% of total development costs). Therefore, CPS are far more difficult to integrate, and testing the hardware is not always practically possible: the final version of the hardware is often available late and testing on the hardware directly can be expensive. A typical approach to dealing with this is to develop hardware proxies, such as prior system hardware versions/simulators and digital twins. However, this approach is flawed and is not sufficient to operate a V&V process that provides high levels of verifiability, trustworthiness, and confidence in the CPS behaviours.
OBJECTIVES
COSMOS overcomes the challenges of developing and evolving high-quality, dependable CPS by employing two key technologies: DevOps and Artificial Intelligence (AI). The collaborative project has:
a) Designed, developed, and validated solutions that continuously improve the overall efficiency/quality of CPS, reducing the number of post-release defects and security vulnerabilities.
b) Developed solutions that enable CPS to autonomously adapt to unexpected run-time behaviours due, for example, to unexpected operating contexts.
c) Investigated the extent to which DevOps concepts can be applied in CPS domains over a diversified set of complex industrial case studies.
The industrial evaluations have demonstrated significant benefits for CPS developers in using the COSMOS tools and technologies for DevOps development processes, along with important business advantages in terms of time and cost savings. Project partners have undertaken substantial dissemination actions to create awareness of the new tools and technologies and have also completed the final dissemination and exploitation planning for the project results with each partner having specific plans extending beyond completion of the project and the continued sustainable evolution of the open source project technologies beyond the end of the project. These actions coupled with the motivations from the improvements demonstrated by the industrial evaluations will ensure substantial and sustainable benefits for the European CPS development community.
Key technology challenges that have been addressed in applying DevOps for CPS include:
a) Simulation and Hardware in-the-Loop (HiL)
b) Representative inputs from signals originating from heterogeneous sensors
c) Run-time verification from distributed logs, data and event streams
d) Security assessment of CPS vulnerabilities
e) Monitoring and self-adaptability of CPS
f) Evolving CPS for faster DevOps cycles
The key industrial impacts targeted by the COSMOS project technologies and innovations are:
i) Reduction the human effort and costs pertaining to the development, verification, validation, and evolution activities of CPS.
ii) Reduction in CPS fault-proneness to security threats, thus contributing to increased trustworthiness and dependability.
iii) Evolving CPS to be able to autonomously adapt (or react) to unexpected behaviour.
COSMOS innovations are driven by development and deployment requirements from five industrial partners providing CPS solutions to key European sectors of Healthcare, Avionics, Automotive, Utilities and Railways. The project has taken specific actions to encourage and support the creation of a European ecosystem for exploiting DevOps for CPS, including open source distribution of project results and an open and transparent industry process for the continued evolution of the COSMOS DevOps technologies.